Sony and Universal Music Group have once again taken Suno to court over alleged copyright infringement, accusing the company of 'model laundering'. The latest legal battle centres on Suno's v6 model, which the labels claim is built on unlicensed music from previous iterations, despite Suno's claims to the contrary.

In its complaint, Sony and UMG argue that training a new model on existing infringing outputs merely 'launder[s]' the copyrighted content, passing its value through successive layers of the model. Jack Brody, CEO of Suno, stated in a previous interview that v6 was ‘trained from the ground up, with a new set of data,’ but the company has been tight-lipped about the specifics of its data sources.

Sony also accuses Suno of using distillation techniques to replicate the results of previous, infringing models. This, according to the complaint, means that even models not directly trained on the labels' recordings benefit from unauthorized data, perpetuating the cycle of copyright infringement.

The ongoing lawsuit highlights the complex challenges in the rapidly evolving field of AI, where data from the past can have lasting impacts on current and future models. For Suno, the battle is far from over, as it faces the task of proving its data sources are legitimate and its models are free from infringement.